| 摘要: |
| 将FMCW雷达检测到的人体生命体征信号，用于预测未来一段时间内人体生命体征信号是否异常，具有明显的应用价值。该方向当前研究主要针对如何进一步降低重构误差、提升生命体征信号的预测精度。为此，本文提出一种自适应变分模态分解?长短期记忆神经网络的生命体征信号预测方法。针对静止状态下的人体，通过雷达采集到的生命体征信号，采用粒子群算法优化变分模态分解VMD的模态分量个数K和惩罚系数α的值，实现自适应选取后用于VMD分解，再将分解后的模态分量进行叠加重构。采用粒子群算法优化长短期记忆网络模型中的网络层数、学习率、正则化系数等3个参数，自适应选取合适的参数组合，将重构后的信号通过优化后的LSTM网络进行预测。实验结果显示本文所提预测方法在10位志愿者的预测结果与原始数据的均方根误差平均值为0.017 188 9，平均绝对误差的平均值为0.007 158，相较于当前其他研究，预测精度上有明显提升。 |
| 关键词: 生命体征信号预测 变分模态分解 长短期记忆递归网络 粒子群算法 |
| DOI：DOI:10.3969/j.issn.1672-2337.2024.01.007 |
| 分类号:TN957.51 |
| 基金项目: |
|
| Vital Sign Signal Prediction Algorithm Based on FMCW Radar |
|
YANG Lu, LEI Yuxiao, YU Xiang
|
|
School of Communications and Information Engineering, Chongqing University of Posts and Telecommunications，Chongqing 400065, China
|
| Abstract: |
| The human vital sign signal detected by FMCW radar can be used to predict whether the human vital sign signal is abnormal in the future period of time, which has obvious application value. The current research in this direction is mainly aimed at how to further reduce the reconstruction error and improve the prediction accuracy of vital sign signal. In this paper, an adaptive variational mode decomposition long short?term memory（LSTM） neural network is proposed to predict vital sign signal. For the human body in a static state, through the vital sign signal collected by radar, the particle swarm optimization algorithm is used to optimize the value of the number of modal components K and penalty coefficient α of the variational mode decomposition VMD, to achieve adaptive selection for VMD decomposition, and then the decomposed modal components are superimposed and reconstructed. The particle swarm optimization algorithm is used to optimize the three parameters of the long short?term memory network model, including the number of network layers, learning rate and regularization coefficient. The appropriate parameter combination is selected adaptively, and the reconstructed signal is predicted through the optimized LSTM network. The experimental results show that the mean square error between the prediction results of 10 volunteers and the original data is 0.017 188 9, and the mean absolute error is 0.007 158. Compared with other current studies, the prediction accuracy is significantly improved. |
| Key words: vital sign signal prediction variational mode decomposition long short⁃term memory neural network particle swarm optimization algorithm(PSO) |
